Claude Mythos 5.2 API pricing: monthly cost calculator

Claude Mythos 5.2 and Fable 5.2 API cost calculator

Updated 2026-11-09. Official Anthropic prices. Costs are our math.

Quick answer

Claude Mythos 5.2 and Fable 5.2 both cost $8 per million input tokens and $40 per million output. A medium month (20M input, 5M output) is $360 at list price, about $235 with 80% of input cached, $180 on the Batch API and $720 in fast mode. GPT-6 Astra costs $450 for the same month and Opus 5.5 $180.

Claude Mythos 5.2 and Fable 5.2: monthly API cost

Workload a month	Standard	80% cached	Batch	Fast mode
Light: side project (2M in, 0.5M out)	$36	$23.52	$18	$72
Medium: small team (20M in, 5M out)	$360	$235	$180	$720
Heavy: production app (200M in, 50M out)	$3,600	$2,352	$1,800	$7,200

Our math at Anthropic list price ($8 input, $40 output per 1M tokens; cache reads $0.20; Batch API 50% off; fast mode $16 and $80). Fable 5.2 costs exactly the same. Cache-write charges ($10 per 1M for 5-minute writes) are left out; Batch cannot combine with fast mode.

Against the other flagships

Model	In / out	Medium month	Heavy month
Claude Mythos 5.2	$8 / $40	$360	$3,600
Claude Fable 5.2	$8 / $40	$360	$3,600
Claude Fable 5.1	$10 / $50	$450	$4,500
Claude Opus 5.5	$4 / $20	$180	$1,800
GPT-6 Astra	$10 / $50	$450	$4,500
Gemini 4 Argon	$2 / $10	$90	$900
Grok 4.7	$2 / $6	$70	$700

Mythos 5.2 and Fable 5.2 share one price list on Anthropic's pricing page. Both are 20% cheaper than the 5.1 generation and 20% cheaper than GPT-6 Astra at list price. Opus 5.5 is half the price. Gemini 4 Argon is cheaper still until Nov 30, after which Google lists $4 input and $20 output.

How to cut the bill

Cache long prompts. A cache read costs $0.20 per million tokens, 2.5% of the input price. At 80% cached input, the medium month falls from $360 to about $235.
Batch anything that can wait. The Batch API halves both input and output.
Skip fast mode unless latency matters. It doubles the price for up to 2.5 times the speed.
US-only inference with inference_geo: "us" adds 10%.

Try your own numbers

Plug Fable 5.2 into our Claude plan vs API calculator or the Max and Ultra version to see when a plan beats paying per token. Mythos uses the same prices.

Verdict

For cost, Opus 5.5 wins among Anthropic models, at half the price. Among the top flagships, Fable 5.2 and Mythos 5.2 beat GPT-6 Astra on list price. With heavy caching, the gap to Astra grows, since Astra's cache read is $1 against $0.20.

Full launch coverage: Claude Fable 5.2 and Mythos 5.2: prices, limits and what changed.

Sources

FAQ

How much does Claude Mythos 5.2 cost per month?

At list price, a medium month of 20M input and 5M output tokens costs $360. With 80% of input cached it is about $235, and on the Batch API $180.

Is Claude Mythos more expensive than Fable 5.2?

No. Anthropic lists the same price for both: $8 input and $40 output per million tokens, $0.20 cache reads and $4 and $20 on the Batch API.

How much is Claude Mythos fast mode?

$16 per million input tokens and $80 per million output tokens, double the standard price, and only for verified organizations.

Is Fable 5.2 cheaper than GPT-6 Astra?

Yes, at list price: $8 and $40 against $10 and $50 per million tokens for Astra prompts up to 272K. A medium month is $360 against $450.

Does prompt caching work on Mythos 5.2?

Yes. Cache reads cost $0.20 per million tokens, 5-minute cache writes $10 and 1-hour cache writes $16.